Currently my company will perform a huge R&R study after repair work to a CMM. This is very time and money consuming. We would like to go to a shortened "repeatablility" test just to verify that the repairs did not affect measurment of product. Any suggestions on sample size, trials, acceptance criteria, or if this is even close to being the "right" thing to do?

Does your repair company also calibrate the CMM after the repair? If so I don't think a gage R&R would be required as long as it was "in calibration" at the time of repair. Was it a repair that affected the dimensional stability of the gage or maybe some other type of repair?

ASD...

Ihas been a number of years since I have dealt with a CMM. When I last did though, if it was an automated system, I would run a ball-bar test 3 to 5 times to determine volumetric accuracy and repeatability. If the results were too varied I would go to a sample of 10 and note any and all variables present and examine for any trends.
If it was a manual system, then a 3 axis linear test covering the measurement range at 1 inch graduations 3 to 5 times was usually sufficient.
I always did this to satisfy my own sense of confidence in the calibration regardless of the documented results I was given.
This has always given me the data I needed to satisfy my functional and system requirements at the time. Would that hold up today? I dunno, but it worked well.

Im curious to read any more posts on this.

Oh yeah, my tests always took less than two hours for the most part, so it wasnt a resource drain at all.

[This message has been edited by Steven Truchon (edited 05 April 2001).]